mm: memcg: fix use after free in mem_cgroup_iter()
authorJing Xia <jing.xia.mail@gmail.com>
Sat, 21 Jul 2018 00:53:48 +0000 (17:53 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Sat, 21 Jul 2018 19:50:46 +0000 (12:50 -0700)
It was reported that a kernel crash happened in mem_cgroup_iter(), which
can be triggered if the legacy cgroup-v1 non-hierarchical mode is used.

Unable to handle kernel paging request at virtual address 6b6b6b6b6b6b8f
......
Call trace:
  mem_cgroup_iter+0x2e0/0x6d4
  shrink_zone+0x8c/0x324
  balance_pgdat+0x450/0x640
  kswapd+0x130/0x4b8
  kthread+0xe8/0xfc
  ret_from_fork+0x10/0x20

  mem_cgroup_iter():
      ......
      if (css_tryget(css))    <-- crash here
    break;
      ......

The crashing reason is that mem_cgroup_iter() uses the memcg object whose
pointer is stored in iter->position, which has been freed before and
filled with POISON_FREE(0x6b).

And the root cause of the use-after-free issue is that
invalidate_reclaim_iterators() fails to reset the value of iter->position
to NULL when the css of the memcg is released in non- hierarchical mode.

Link: http://lkml.kernel.org/r/1531994807-25639-1-git-send-email-jing.xia@unisoc.com
Fixes: 6df38689e0e9 ("mm: memcontrol: fix possible memcg leak due to interrupted reclaim")
Signed-off-by: Jing Xia <jing.xia.mail@gmail.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: <chunyan.zhang@unisoc.com>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/memcontrol.c

index e6f0d5ef320aa65d2b65ceed4b202021a84fd49b..8c0280b3143ee241053ccdf0eb31e9790edfef0c 100644 (file)
@@ -850,7 +850,7 @@ static void invalidate_reclaim_iterators(struct mem_cgroup *dead_memcg)
        int nid;
        int i;
 
-       while ((memcg = parent_mem_cgroup(memcg))) {
+       for (; memcg; memcg = parent_mem_cgroup(memcg)) {
                for_each_node(nid) {
                        mz = mem_cgroup_nodeinfo(memcg, nid);
                        for (i = 0; i <= DEF_PRIORITY; i++) {